独白：苍白骑士降临

这是iHeart播客。

This is an iHeart podcast.

保证真人主持。

Guaranteed human.

经营企业却没想过做播客？

Run a business and not thinking about podcasting?

再想想吧。

Think again.

现在听播客的美国人，比用Spotify和Pandora听广告支持流媒体音乐的人还多。

More Americans listen to podcasts than ad supported streaming music from Spotify and Pandora.

作为排名第一的播客平台，iHeart的规模是第二名和第三名加起来的两倍。

And as the number one podcaster, iHeart's twice as large as the next two combined.

来了解播客如何助力你的业务。

Learn how podcasting can help your business.

拨打844844。

Call 844844

嘿，波特兰迪亚的粉丝们。

Hey, Portlandia fans.

这里是凯莉·布朗斯坦和弗雷德·阿米森。

Carrie Brownstein and Fred Armisen here.

九十年代的梦想以播客的形式复活了。

The dream of the nineties is alive in podcast form.

我们要推出全新的播客《波德兰迪亚 AO 重看》，在这里我们会一起重温每一集，拆解其中的小品段子，深入探讨我们那些标志性的角色，并揭开这部剧制作背后的秘密。

We're launching Podlandia AO Rewatch, our brand new podcast where we revisit every episode of Portlandia together, breaking down sketches, going deep on our iconic characters, and pulling back the curtain on how it all got made.

我们还会邀请到那些让这一切成为现实的人，包括客串明星、合作者和朋友们，比如导演乔纳森·克莱塞尔、市长本人凯尔·麦克劳克林、传奇音乐家艾米·曼等等。

And we'll also be joined by the people who helped bring it all to life, Guest stars, collaborators, and friends, including director Jonathan Kreisel, the mayor himself, Kyle McLaughlin, legendary musician Amy Mann, and many more.

凯尔这次可是全力以赴啊。

Kyle is going for it here.

你不仅即兴创作了台词，还即兴唱了一首歌。我当时还以为……

You fully improvised not just words, but but a song about Well, I thought you were

你要写一首歌呢。

all gonna write a song.

我记得你，是的。

I remember you Yeah.

当时是这么想的。

Thinking that.

听一听

Listen

Podlandia AAO 重看播客，在 iHeartRadio 应用、Apple Podcasts 或你收听播客的任何平台都能找到。

to Podlandia, AAO rewatch on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.

我是切尔西·哈姆勒，来自《亲爱的切尔西》节目。

This is Chelsea Hamler from Dear Chelsea.

每周新闻越来越糟，世界越来越疯狂，而雅玛内卡会告诉那些该负责的人：问题就出在你们身上。

Every week, the news gets worse, the world gets crazier, and Yamaneka is here to tell whoever's responsible, you're the problem.

你知道吗，我刚发现西德尼·斯威尼是谁。

Do you know I just found out who Sydney Sweeney was?

如果他有一大群女人，那我也应该有一大群男人。

If he got a bunch of women, then I should have a bunch of men.

做得更好点，或者少做点，这样我就不用做那么多了。

Do better or do less so I don't have to do so much.

我是雅曼尼卡，先走了。

I'm Yamanika, and I'm out.

欢迎收听《你的问题》节目，雅曼尼卡主持，

Listen to You're the Problem with Yamanika on

在iHeartRadio应用、Apple Podcasts或任何你收听播客的平台都能找到。

the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.

我是尼克·塔图罗。

I'm Nick Taturo.

你可能从《纽约重案组》、《最长的一码》或斯派克·李的《黑色党徒》里认识我。

You probably know me from NYPD Blue, The Longest Yard, or Spike Lee's Black Plansman.

在我的新播客《与尼克·塔图罗一起传递快乐》里，我会给新嘉宾送披萨。

And on my new podcast, Delivering Happiness with Nick Taturo, I deliver pizza to a new guest.

我和很多人分享过披萨，从塞斯·罗林斯——你拿我的腰带干什么？

I've shared a slice with everyone from Seth Rollins What are you doing with my belt?

到比尔·伯尔。

To Bill Burr.

我从来没见过有人像我一样疯得这么彻底。

I don't think I've ever met somebody so exactly out of their mind as I am.

现在还有更多重量级嘉宾要来，包括了不起的约翰·塔图罗。

And now we even have more great guests coming up, including the great John Taturo.

这就是幸福。

It's called happiness.

传递幸福。

Delivering happiness.

还有更多，更多。

And many, many more.

打开免费的iHeartRadio应用，搜索《与尼克·塔图罗一起传递幸福》，现在就听。

Open your free iHeartRadio app, search delivering happiness with Nick Taturo, and listen now.

Callzone Media。

Callzone Media.

我一看，只见一匹 pale horse，骑者的名字叫微笑之人。

And I looked and behold, a pale horse and the rider's name was the smiling man.

这是你每周的 Better Offline 独白，我是主持人 Ed Zitron。

This is your weekly Better Offline monologue and I'm your host Ed Zitron.

这

This

周，我发布了我有史以来最重要的 newsletter 之一，叫做“AI 需求泡沫”，它汇集了富国银行、巴克莱和瑞银的多份独立金融分析师报告，这些报告直接估计微软、谷歌和亚马逊的 AI 收入中有 70% 或更多来自 OpenAI 或 Anthropic。

week, I put out one of my most consequential newsletters of all time called the AI demand bubble, pulling together multiple distinct financial analyst notes from Wells Fargo, Barclays and UBS that directly estimated that 70% or more of the AI revenues of Microsoft, Google and Amazon were from either OpenAI or Anthropic.

明确地说，瑞银估计明年 Anthropic 和 OpenAI 的计算支出将占谷歌云总收入的 48%，这意味着它们可能会占到 70% 以上。

To be clear, UBS estimated the next year Anthropic and OpenAI's compute spend will be 48% of all Google Cloud revenues, which means that they will likely account for even more than 70%.

但我想公平一点。

But I wanted to be fair.

这件事既是个巨大的麻烦，也是我知道会惹恼很多人的故事，因为它影响深远。

This was both a colossal pain in the ass and the story that I knew would piss off a lot of people because of its huge ramifications.

结果，我遭到了很多人的反驳，他们要么完全否认这一点，要么说，实际上，这两个不可持续的公司占据着那些花了超过一万亿美元的公司的大部分 AI 收入，这是好事。

As a result, I've had a ton of pushback from people that either outright deny that this is the case or say, actually, it's good the two unsustainable companies are the majority of AI revenues for companies that spent over a trillion dollars.

其实，这太棒了。

Actually, that's great.

总之，二十四小时后，彭博社报道了一则消息，根据OpenAI对微软2026财年收入的241亿美元贡献以及此前的声明估算，仅OpenAI就占到了微软全年AI收入的70%以上。

Anyway, twenty four hours later, Bloomberg ran a story estimating based on OpenAI's $24,100,000,000 contribution to Microsoft's fiscal year twenty twenty six revenues and previous statements, the OpenAI alone contributed to 70% or more of Microsoft's AI revenues for the year.

作为背景参考，自2022年初以来，微软在资本支出上已经投入了261.3亿美元。

For some context, Microsoft has spent $26,130,000,000 in capital expenditures since the beginning of 2022.

除了那种刚写完一篇七千字、基于深度研究和硬数据的论文，结果不到一天就被验证的得意感之外，这个故事也证实了我对AI行业现状以及AI算力真实需求最大的担忧。

And other than getting to have the swagger of someone who just laid out a 7,000 word thesis founded in deep research and hard numbers only to get their work validated in less than a day, this story also confirms my greatest fears about the state of the AI industry and the actual demand for AI compute.

微软已经砸了超过两千亿美元，这还没算上投给OpenAI的一百三十亿，以及为了留住客户、按成本价运营服务所亏掉的钱——至少在2024年是这样，甚至可能更久。结果呢，这个客户只占它年收入的7%，而且要是没有风投源源不断的资金注入，它根本活不下去。

Microsoft has spent over $200,000,000,000, and that's not including the 13,000,000,000 invested in OpenAI or the losses it's incurred running their services at cost, at least in 2024, but potentially beyond, to create a customer that is now worth 7% of its annual revenue that cannot sustain its existence without near constant flows of venture capital funding.

微软2026财年的同比增长率大概是17.7%。但如果你把OpenAI贡献的那两百四十一亿美元收入（来自算力支出和收入分成）剔除掉，微软的实际同比增长只有9.24%左右，这是自2023财年以来的最低增速，当时它的增幅才7%。

Microsoft's year over year growth for fiscal year twenty twenty six was about 17.7%, But when you remove OpenAI's $24,100,000,000 in revenue, which comes from OpenAI's compute spend and revenue share, Microsoft's year over year growth was more like 9.24%, the lowest it would be since fiscal year twenty twenty three when it only grew by 7%.

不过有个很大的区别：整个2023财年的资本支出是三百一十八亿美元，而微软在2026财年第四季度就花了三百五十八亿美元。那一年它总共花了一千一百五十亿美元，天杀的，而且它还计划在2027财年（从2026年7月1日开始）再投入一千七百五十亿美元的资本支出。

But there is one big difference and that, well, capital expenditures in the entire fiscal year 2023 were $31,800,000,000 and, Microsoft spent $35,800,000,000 in the fourth quarter of the fiscal year of 2026, a year when it spent $115,000,000,000 goddamn dollars and it wants to spend another $175,000,000,000 in capex in fiscal year twenty twenty seven, which begins 07/01/2026.

挺让人烦的，但我们继续往下说。

It's annoying but we're moving on.

哦对了，OpenAI，可以说是微软最大的客户，上个财年几乎贡献了微软一半的增长，它快没钱了。

Oh, and OpenAI, arguably Microsoft's largest customer and responsible for nearly half of its growth in the last fiscal year, is gonna run out of money.

而且由于OpenAI占了微软AI收入的70%，很明显没有任何其他大客户存在，能把这些基础设施全部消化掉。

And by nature of it being 70% of Microsoft's AI revenues, it's very obvious that there are no other large customers that exist that could possibly take up all of this infrastructure.

当然，除了那个古怪的Wario Amadez Anthropic，它跟OpenAI处境一模一样，只是声音更烦人一点，因为我们快接近指数级了。

Other than, of course, weird Wario Amadez Anthropic who is in exactly the same position as OpenAI, but with a slightly more annoying voice because we're near the exponential.

这也意味着，现在所有未来的收入指引和分析师预期，都是基于一个明确假设：Anthropic和OpenAI都会有需求，并且有能力以越来越高的价格，购买越来越大规模的计算能力。

This also means that any future revenue guidance and analyst expectations are now set under the explicit assumption that Anthropic and OpenAI will both have the need for and be able to pay for increasingly larger levels of compute at increasingly higher prices.

根据我之前报道的分析师报告，那些内容都在通讯里，至少引用了，瑞银、富国、巴克莱都预期，未来几年谷歌、亚马逊和微软的大部分收入会来自OpenAI和Anthropic，而且他们似乎一点都没有偏离这个观点的意思。

Based on the analyst notes I reported on, which are in the newsletter, or at least quotes, UBS, Wells Fargo, Barclays all have the expectation that much of Google, Amazon, and Microsoft's revenues are gonna come from OpenAI and Anthropic for the next few years, and they don't seem to be diverging from that point in any any way.

大家都觉得收入就得靠这个来源。

Everyone just thinks that this is where it's gonna come from.

如果这些收入因为破产、需求不足或数据中心没建好而落空，也就是说他们收不到钱，那就会惹恼华尔街，因为华尔街被洗脑了，以为这三家公司是因为AI服务的多元化需求才增长的，而不是把一万亿美元砸进水里，大部分只是自己在喂自己钱。

Any failure for these revenues to appear out of insolvency, lack of demand, or unbuilt data center capacity, meaning they can't get paid, will piss off a Wall Street that's been brainwashed into believing that these three companies are growing because of diverse demand for AI services rather than having pissed a trillion dollars up the wall to mostly to just feed themselves money.

我不太相信市场真的能理解彭博说的这些。

I am not confident that the market will actually understand what Bloomberg has said.

我现在看的时候，微软涨了2%。

As I'm looking now, Microsoft is up 2%.

我的天，真是见鬼了。

Jesus fucking Christ.

所以让我给你说清楚。

So let me spell it out for you.

微软AI收入的70%要么来自OpenAI收入的分成，要么来自OpenAI租用Azure算力的支出。

70% of Microsoft's AI revenues are from either its revenue share of OpenAI's money or OpenAI spending money to rent outcompute from Azure.

这意味着微软现在依赖OpenAI来推动未来的收入增长，因为OpenAI现在占微软年收入的7%以上，而除了微软365 Copilot每年几十亿美元的收入外，似乎没有其他显著的贡献来源，而Copilot这个产品主要是靠像《公主与青蛙》开头那个Facilier医生一样，骗人签合同来卖的。

This means that Microsoft is now dependent on OpenAI for future revenue growth as the company now represents more than 7% of its annual revenues and there does not appear to be significant contribution from anything outside of single digit billions of annual revenue from Microsoft three sixty five Copilot, a product mostly sold by tricking people into signing contracts like Doctor.

Facilier医生。

Facilier in the beginning of Princess and the Frog.

微软曾以为对OpenAI的投资会创造出一个独立的超速增长引擎。

Microsoft believed it would create an independent ultra growth engine from its investments in OpenAI.

结果，它创造了一个寄生虫。

Instead, it has created a parasite.

AI收入里有70%都来自OpenAI，这说明萨提亚·纳德拉的帝国已经失败了。

At 70% of AI revenues, OpenAI represents the failure of Satya Nadella's empire.

微软在基础设施上砸了2600亿美元，但这些需求除了两家公司之外根本不存在，而那两家公司每年都要烧掉几百亿美元。

Microsoft has spent $260,000,000,000 on infrastructure for demand that does not exist outside of two different companies that burn tens of billions of dollars a year.

更糟糕的是，OpenAI和Anthropic为了跟上分析师的预期，每年都必须大幅增加算力支出。这意味着它们需要越来越多的钱，所以投资微软的人，其实就是在赌OpenAI和Anthropic还能撑多久、还能融到多少资，直到它们把能用的资金渠道全部耗尽。

And to make matters worse, OpenAI and Anthropic must grow their compute spend aggressively every single year to keep up with the expectations of analysts, which will mean it will need more and more money, which means anyone investing in Microsoft is effectively gambling on how long OpenAI and Anthropic can continue to raise money before they simply exhaust every available form of capital.

这也说明，AI并没有真正的市场需求，至少跟那些巨额资本开支，或者人们没完没了地吹嘘AI将如何改变世界相比，完全不成比例。

This also means that there's not really real demand for AI, at least not remotely commensurate with the capital expenditures or all of the endless blathering on about how much the world is changing because of AI.

微软有数以万计的Azure和软件销售人员，在构建大规模基础设施方面拥有仅次于AWS的专业能力，并且拥有一位仅次于甲骨文的、极其冷酷无情的首席财务官。

Microsoft has tens of thousands of Azure and software salespeople, the single most expertise in building large scale infrastructure outside of AWS, and the single most ruthless CFO outside of Oracle.

如果真的有实际需求、真实的利润空间和真正的客户，他们早就把服务卖给任何其他公司了。

If there were actual demand, actual margins, actual customers, they would sell to literally anybody else.

他们会把算力卖给除了OpenAI以外的任何其他人。

They would sell that compute to literally anyone else other than OpenAI.

哪怕对方是加格尔莫（Gargamel）这种角色，他们也会卖。

They would sell it to Gargamel.

微软宁愿帮加格梅尔杀掉每一个蓝精灵，也再也不想跟该死的萨姆·奥特曼或达里奥·阿莫迪说一句话了。

Microsoft would happily help Gargamel kill every single smurf rather than talk to Sam fucking Altman or Dario Amade ever again.

如果AI算力或AI软件真的有实质性的需求，作为全球最大的这两类供应商，微软理应成为这种需求的代表。

If there were meaningful demand for AI compute or AI software, Microsoft would be representative of it as the largest vendor in the world of both.

它70%的AI收入和7%的年度总收入都来自同一家公司——这家公司今年三月刚融了1220亿美元，还因为顾问认为拿不到万亿美元估值而推迟了IPO。

70% of its AI revenues and 7% of its entire annual revenues come from a company that just had to raise a $122,000,000,000 in March and then delay its IPO because its advisers didn't think it would get a trillion dollar valuation.

对于超过300吉瓦的数据中心容量和规划来说，需求根本不存在。当然，除了OpenAI和Anthropic，它们至今还没证明任何可能实现盈利的方法，过去七个月里仅仅为了维持运营就合计融资超过了2000亿美元。

The demand is not there for the over 300 gigawatts of data center capacity and planning outside, of course, of OpenAI and Anthropic, who still to this day have yet to prove any possible way of becoming profitable and have had to raise over a combined $200,000,000,000 in the last seven months just to keep the lights on.

说清楚点，萨姆·奥特曼也签下了超过7500亿美元的承诺。这意味着谷歌、亚马逊、微软、CoreWeave、Cerberus、Iron、Nebius、Lambda以及其他所有相关交易对手的未来收入增长，都取决于这个黏糊糊的萨姆·奥特曼能否在接下来三年内筹集到比全球风险投资过去三年总和还要多的资金。

To be clear, Sam Altman has also signed up for over $750,000,000,000 worth of commitments, which means that Google, Amazon, Microsoft, CoreWeave, Cerberus, Iron, Nebius, Lambda, and any other associated counterparties' future revenue growth is dependent on clammy Sam Altman's ability to raise more than the combined sum of global venture capital in the last year in the next three years.

除此之外，达里奥·阿马德伊领导的Anthropic也做出了超过3000亿美元的承诺，这还不包括那笔由Apollo支持的350亿美元私募信贷交易——根据该交易，谷歌会将TPU出售给一个特殊目的载体（SPV），然后把这些TPU安装在谷歌的数据中心里，再由Anthropic向谷歌租用。

On top of that, Dario Amadeus Anthropic has made over $300,000,000,000 worth of commitments on top of a $35,000,000,000 Apollo backed private credit deal where Google will sell its TPUs to an SPV that would then install them in Google data centers for Anthropic to then rent from Google.

这也太蠢了。

It's so stupid.

这简直是一座疯狂的纪念碑，标志着这个狗屎时代里那些绝望又狂热的资本主义创新。

It's an insane monument to the desperate manic capitalistic innovations that crown this wretched era of dog shit.

做企业还不考虑播客？

Run a business and not thinking about podcasting?

再好好想想吧。

Think again.

听播客的美国人数比听 iHeartRadio 的还多。

More Americans listen to podcasts than ad Think iHeart.

流媒体、广播和播客。

Streaming, radio, and podcasting.

请到 iheartadvertising.com 看看。

Let us show you at iheartadvertising.com.

网址就是 iheartadvertising.com。

That's iheartadvertising.com.

嘿，《波特兰迪亚》的粉丝们。

Hey, Portlandia fans.

这里是凯莉·布朗斯坦和弗雷德·阿米森。

Carrie Brownstein and Fred Armisen here.

你认识我们，或者说，你认识他们。

You know us or rather you know them.

托尼和坎迪斯，妮娜和兰斯，斯派克，还有，没错，那只鸡。

Tony and Candice, Nina and Lance, Spike, and, yes, the chicken.

这些年来我们演过很多经典角色，但今天我们要以真实身份出现，来跟你聊聊我们的全新播客《波特兰迪亚重看》。

We've played a lot of iconic characters over the years, but today we're showing up as ourselves to tell you about Podlandia AO Rewatch, our brand new podcast.

每周我们都会从头回顾一集《波特兰迪亚》，拆解其中的小品，探索我们最经典角色的背景故事，重温你熟悉又喜爱的波特兰取景地，并且坦诚分享我们的创作过程。

Each week, we'll revisit an episode of Portlandia from the very beginning, breaking down the sketches, exploring the backstories of our most iconic characters, revisiting the Portland locations you know and love, and opening up about our creative process.

这一切到底是怎么做出来的？

How did any of this get made?

为什么我们会觉得那是个好主意？

Why do we think that was a good idea?

我们准备好聊这些了。

We're ready to talk about it.

而且我们还会邀请那些帮助这一切成真的人一起加入。

And we'll also be joined by the people who helped bring it all to life.

还有客串明星、合作伙伴们以及朋友们，比如导演乔纳森·克莱塞尔、市长本人凯尔·麦克劳克林、传奇音乐人艾米·曼，以及更多嘉宾。

Guest stars, collaborators, and friends, including director Jonathan Kreisel, the mayor himself, Kyle McLaughlin, legendary musician Amy Mann, and many more.

凯尔这次可是豁出去了。

Kyle is going for it here.

你完全是即兴发挥的，不只是台词，连那首歌都是现编的。

You fully improvised not of just words, but a song about it.

他本来打算写歌的，我还以为你们都要写一首呢。

He was gonna write I thought you're all gonna write a song.

我记得你当时是这么想的。

I remember you thinking that.

去听听《Podlandia》吧。

Listen to Podlandia.

哎哟。

Ayo.

在 iHeartRadio App、Apple Podcasts，或者你平时听播客的任何平台都能找到重看版。

Rewatch on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.

嘿。

Hey.

我是海斯·达文波特。

This is Hayes Davenport.

还有肖恩·克莱门茨。

And Sean Clements.

我们主持一个播客叫《好莱坞手册》。

We host the podcast Hollywood Handbook.

每一集，我们都试图帮助嘉宾提升事业，但主要关注的是他们怎么帮我们提升事业。

Every episode, we're trying to help our guests improve their careers by mainly focusing on how they can help us improve our careers.

这个节目出了名的容易上手，所以你可以随便挑我们650集里的一集听，都能明白在讲什么。

The show is famously super accessible, so you can easily jump into any of our 650 episodes and understand what's going on.

我们推荐几集最近和本·斯蒂勒一起录的。

We recommend some of our recent episodes with Ben Stiller.

其实现在对于什么是电影的定义也已经有点变了。

The definition of what is a movie has kinda changed anyway too.

对吧？

Right?

你觉得现在算什么？

What do you think it is now?

就是图像和话语，用某种技术捕捉下来。

It's images and words being spoken captured by some sort of technology.

丹尼·麦克布莱德

Danny McBride

他超爱珍珠奶茶。

He loves boba.

哦，那太好了。

Oh, that's fantastic.

而且你是自己做的。

And you make it on you make your own.

真不错。

That's great.

对。

Mhmm.

我管他叫芋头·里德。

I call him Taro Reed.

这名字不错。

It's a good name.

他们都是以《美国派》的演员名字命名的。

They're all named after cast members of American Pie.

是的。

Yeah.

哦，明白了。

Oh, gotcha.

我不知道她有没有出演，但我管他叫荔枝·索比斯基。

I call I don't I never looked up if she was in this, but I call him Lychee Sobieski.

对。

Yeah.

我觉得她没演过《美国派》，不过我

I don't think she was in American Pie, but I

还没去查过。

Haven't haven't looked it up.

确实没查过。

Haven't looked it up.

对。

Yeah.

她当时

She was

挺有名的。对。

prominent around Yeah.

还有

Same

玛丽·斯汀伯根。

Then Mary Steenburgen.

你们干嘛跟我说话？

Why are you guys talking to me?

我是说，我懂这部电影。

Like, I understand the movie?

别把我当傻子，以为我不明白你们那点小心思。

So down to me that you think I don't understand your little play play game.

我就是真的懂了。

I just really I get it.

我是想帮你把基础打牢，让你能顺利上手。

Trying to set you up for success here.

你可以在 iHeartRadio 应用、Apple Podcasts，或者你平时听播客的任何平台上收听《好莱坞手册》。

Listen to Hollywood Handbook on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.

在 Solita 这个节目里，我们会分享女性独自旅行时那些真实又充满挑战的经历。

On Solita, we share the messy stories of traveling alone as a woman.

光靠多邻国（Duolingo）学的那点东西，真到了关键时刻可派不上用场。

Your Duolingo is not going to come back to you when a man is inside of you.

哦，不。

Oh, no.

我是朱莉·皮涅罗，我独自旅行，因为这是一个难得的空间，我可以不用先问别人就说是。

I'm Julie Pinheiro, and I travel by myself because it's a rare space where I can say yes without asking anyone else first.

我的使命是重新定义“solita”这个词，把怜悯变成可能性。

I'm on a mission to reclaim the word solita, trading the pity for possibility.

每次我想独处的时候，总会遇到人，他们就说，你对我们笑了。

Every time I tried to be alone, I kept meeting people, and they were like, you smiled at us.

这附近没多少人会笑。

Not a lot of people smile around here.

我可以等四个

I can wait four

小时等下一班车，或者这个陌生男人让我搭他的摩托车。

hours for the next bus, or this random dude is offering me a ride on his motorcycle.

只有当你独自一人时，你才能最真实地感受这个世界，而不是别人灌输给你的那些谎言。

It's when you're alone that you're most receptive to the world as it is and not the lies you're sold about it.

所以，无论你是独自旅行的老手，还是紧张得不敢预订第一次旅行，我希望你收听iHeartRadio应用、Apple Podcasts或任何你获取播客的平台上的Solita。

So whether you're a solo travel veteran or you're too nervous to book your first trip, I hope you listen to Solita on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.

需求并不存在。

The demand isn't there.

我知道你们有些人可能会给我发邮件说，嗯，Anthropic和OpenAI确实有需求。

And some of you might I know I'm gonna get an email saying, well, demand is real from Anthropic and OpenAI.

对。

Yeah.

他们以1美元的价格卖20或40美元的东西。

They're selling $20 or $40 for a dollar.

我想确实有一些需求。

I I imagine there's some demand.

每个人都在催促你到处使用AI。

Everyone's screaming at you to use AI everywhere.

哇。

Wow.

这种需求是来自社会压力的，不是真正的价值或实用性。

That's demand that comes from social pressure, not actual value or utility.

而且说实话，这种需求根本不存在。

And really, that demand isn't there.

如果微软都没看到这种需求，那其他人更看不到了。

If Microsoft isn't seeing it, nobody is seeing it.

任何新云厂商还能活着，唯一的原因就是英伟达的循环融资把戏、评级机构的失职、OpenAI和Anthropic，以及马克·扎克伯格搞的那些乱七八糟的东西。

The only reason that any Neo Cloud is still alive is NVIDIA's circular financing rituals, failure of ratings agencies, OpenAI and Anthropic, and whatever weirdo bullshit it is that Mark Zuckerberg is up to.

Metis计算能力第一个也是唯一一个客户是Anthropic，这背后是有原因的。

There's a reason that the first and only story about a customer for Metis compute capacity was Anthropic.

没有其他人了。

There's nobody else.

除了Anthropic，当然还有OpenAI，或者那些超大规模云厂商，他们在不断吸走AI实验室的风险投资资金来维持自己的增长，每次融资都多要一点，估值更高一点，每次都要求更多的钱，希望最终能把这些债务甩到公开市场上，指望能永远骗投资者来支持他们的交易。

Nobody else other than Anthropic, of course, OpenAI, or the hyperscalers that are siphoning off what remains of the AI labs venture capital funding to sustain their own growth rates, every time raising a little bit more and a little bit higher at a higher valuation, each time demanding more and more money, hoping to be able to dump it onto the public markets in the hope that they can perpetually con investors into backing their deals.

微软、谷歌和亚马逊之所以对AI收入这么含糊其辞，是因为他们知道这些收入跟他们的资本支出相比简直微不足道，而且几乎完全依赖两家公司，超大规模云厂商必须让这两家公司活着，这样他们才能付钱给他们。

The reason that Microsoft and Google and Amazon have been so squirrelly about their AI revenues is that they know that they're pathetic compared to their capex and near entirely dependent on two companies that hyperscalers must keep alive so that they can they can pay them.

他们能拿到的钱，其实都是超大规模云厂商给的钱。

They can get paid by them with the money that the hyperscalers handed to them.

这情况可不妙。

It's not good.

光2026年一年，谷歌和亚马逊就向Anthropic联合投资了150亿美元，而亚马逊对OpenAI的投资更是高达500亿美元。

And in 2026 alone, Google and Amazon have invested a combined $15,000,000,000 in Anthropic, and Amazon's invested a combined $50,000,000,000 in OpenAI.

更离谱的是，这350亿美元里有很多是附带条件的，得等到IPO或者实现AGI才能兑现。

What's crazy about that was $35,000,000,000 of that was contingent on the IPO or AGI.

这两件事都没发生，这强烈表明亚马逊其实心里清楚，他们得先付钱给OpenAI，这样OpenAI才有能力反过来付钱给他们。

Neither of those happened, which heavily suggests that Amazon kinda knew that they needed to pay OpenAI so that OpenAI could pay them.

顺便说一句，当你没有其他选择、没有别的靠谱客户，而且核心业务的增长速度根本没法让华尔街满意时，你只能这么干。

And by the way, this is what you do when you have no other options, no other viable customers, and your core business is not growing at the rate necessary to appease Wall Street.

这事儿明摆着放在那儿好几年了。

It's been sitting in plain sight for years.

超大规模云厂商过度建设了算力容量，还资助了他们最大的客户——实际上是在孵化这些客户。他们以为，既拥有自己的模型又拥有底层算力，就能像开了无限刷钱外挂一样爽。每个超大规模厂商都觉得自己同时握有下一代云计算基础设施和软件平台的重磅机会，同时还通过控股在两大巨头身上制造依赖关系。

Hyperscalers overbuilt capacity, funded their largest customers, incubating them actually, and believed that both the opportunity of owning their models and the underlying compute capacity would be some sort of infinite money glitch, with each hyperscaler owning the equivalent of both the next big cloud infrastructure and cloud software plays while also owning controlling stakes and creating dependency from the two largest players in both.

挺简单的。

Pretty easy.

对吧？

Right?

其他建AI数据中心的人，都相信微软、谷歌和亚马逊不会蠢到在没有任何铁定保证的情况下做这一切，保证这些投入一定能变成他们承诺的那样，保证资本支出肯定值得，有足够的市场需求支撑，而且他们不会在没有这种确定性的时候继续砸钱。

Anyone else building AI data centers believed that Microsoft, Google, and Amazon would never be as foolish as to do all of this without a cast iron guarantee that it would turn into everything they'd promised, and that the capital expenditures would be, for certain, worth it and have the demand to back them and that they would not continue spending without that certainty.

现实是，大型云服务商犯了一个巨大的错误，既高估了AI的能力，高估了OpenAI和Anthropic能多快增长，也低估了这些业务在扩张过程中会变得多糟糕。

The reality is that hyperscalers made a huge mistake, both in what AI could do, how fast OpenAI and Anthropic could grow, and in underestimating how bad these businesses would get as they grew.

事后看来，这也是一目了然的。

It was also obvious in retrospect.

从一开始，大型云服务商就不愿公布他们的AI收入，即使公布了，他们也用不具体的时期来包装欺骗性的年化运行率。

From day one hyperscalers refused to share their AI revenues, and when they did so they used deceptive run rates based on non specific periods.

我说了好几年了。

I've said this for years.

如果这些公司有多元化且可观的AI收入，他们肯定会清清楚楚、大张旗鼓地到处宣传。

If these companies had diverse and meaningful AI revenues, they would be screaming about them from the rooftops with clarity and vigor.

他们会按业务板块告诉你AI对收入的贡献。

They would tell you segment by segment AI specific contribution to revenue.

他们会把成本单独列出来。

They break out the costs.

他们会用大号字体写在投资者演示文稿上。

They would have them in big font on investor presentations.

但谷歌却告诉我们, 它每月处理3.2千万亿个token。

Instead, Google is telling us it processed 3.2 quadrillion tokens a month.

谁在乎这个啊, 桑达尔?

Who gives a fuck, Sunda?

你今年打算花1800亿美元, 我们心里都清楚, 如果你有能勉强支撑这么大支出的收入, 你早就把它们列出来了。

You're planning to spend a $180,000,000,000 this year, and we both know that if you had the revenues that would remotely justify spending that much, you'd include them.

你是不是犯了大错, 桑达尔?

Did you make a big mistake, Sundar?

你可以告诉我, 老兄。

You can tell me, buddy.

我不会生你的气。

I won't be mad at you.

但投资者会。

Investors will be, though.

他们会非常愤怒。

They're they're gonna be furious.

但往好处想。

But look at the bright side.

至少你不用再听那个无聊的德米斯·哈萨比斯唠叨个不停，说什么，哦，要是电脑有最聪明的大脑会怎样？

At least you won't have to hear from that boring oath, Demis Hassabis, and hearing him prattle on and on about like, oh, what if the computer had the most brilliant mind?

要是我们二十一分钟内就有AGI呢？

What if we had AGI in twenty one minutes?

现在他要走了，我猜他会花更多时间陪他的钱。

Now he's gonna go off and I assume spend more time with his money.

总之，总有一天大家都会装作非常惊讶，说根本没想到会这样。

Anyway, one day everybody is gonna act super surprised and say there was no way to see this coming.

要知道，这些迹象已经存在好几年了。

Know that the signs were there for years.

它们一直没变过。

They've stayed the same.

唯一变的是OpenAI和Anthropic的估值被吹得越来越高，还有不断升级的资本开支，把半导体泡沫越吹越大，这个泡沫之所以爆掉，是因为没人想到超大规模计算公司会白白烧掉一万亿美元。

The only thing that's changed is the inflated value of OpenAI and Anthropic and the ever escalating capex fueling a semiconductor bubble that erupted because nobody thought hyperscalers would ever waste $1,000,000,000,000.

总有一天，资助这两家公司的钱会烧完，到那时，超大规模计算公司就再也没法给自己喂钱了。

At some point, the money runs out to fund these two companies, at which point hyperscalers will have no way to further feed themselves money.

我以前就说过，AI泡沫是一场心理战、一场闹剧、一场金融危机，也是那些掌管着大部分经济的商业蠢货们摘下面具的时刻。

As I've said before, the AI bubble is a psyop, a melodrama, a financial crisis, and a mask off moment for the business idiots that run the vast majority of the economy.

这是历史上最奇怪的时刻之一，而且随着事情越来越 desperate，只会变得更奇怪。

It is one of the strangest moments in history, and it will only get stranger as things get more desperate.

我们现在看到的是，当腐朽的经济要求我们把世界的资源——无论是物理上的还是财政上的——都推到极限，去追求永恒的增长时，会发生什么；以及当我们发现这些极限到底是什么时，又会发生什么。

We are seeing what happens when the rot economy demands that we push the world's resources to their very limits, both physically and fiscally, in pursuit of eternal growth, and what happens when we discover what those limits actually are.

这一切从来就无关技术，也无关什么好生意，唯一的目的就是为资本创造下一个增长引擎。

None of this was ever about technology or good business or anything other than creating the next growth engine for capital.

相反，人工智能已经成了文化、金融和生态上的寄生虫，在它崩塌之前会尽可能消耗一切。

Instead, AI has become a parasitic presence culturally, financially, and ecologically, and it will exhaust as much as it can in its collapse.

这根本没法救，问题不在于钱不存在，而在于钱不是自然产生的。

Nothing about this can be bailed out because the problem isn't that the money doesn't exist, just that it doesn't exist naturally.

一旦谷歌、微软、Meta、亚马逊或甲骨文的增长彻底放缓，任何救助都无法让它们重新增长。

A bailout cannot make growth restart for Google, Microsoft, Meta, Amazon or Oracle once it fully slows.

而且，考虑到他们已经在AI上砸了超过一万亿美元的资本支出，我觉得他们肯定明白，如果不搞AI——我引用一下——增长要么已经没了，要么很快就会死，而搞AI主要就是指望着Anthropic和OpenAI每个季度、永远地付给他们越来越多的钱。

And based on the fact they've sunk more than a trillion dollars of capex into AI, I think they must clearly see that growth is already or will die very soon without, and I quote, doing AI, which mostly means hoping Anthropic and OpenAI pay them ever increasing amounts of money every quarter forever.

美国政府没法撑住一个接受了AI增长有限的股市，而且除了半导体公司和两家受补贴的AI实验室之外，最好的情况也就是几百亿美元的需求。

The US government cannot prop up a stock market that accepts that AI's growth is finite and, outside of semiconductor companies and two subsidized AI labs, is tens of billions of dollars of demand in the best case scenario.

我知道很容易陷入一种心态，觉得什么都能被救，坏人总能赢，因为我们太习惯看到那些腐败的坏蛋逃脱责任了。

I realize how easy it is to fall into the mindset that everything gets bailed out, that the evil people always win, because we're so used to seeing these evil corrupt people escape accountability.

但在这个情况下，虽然我觉得没人会进监狱，但我觉得这些公司最终会耗尽资源，然后退出。

In this case, while I don't think anyone's going to prison, I think these companies simply exhaust their resources at some point and drop off.

我觉得超大规模增长的时代要结束了，他们之所以拼命挤进那个没有IT负载、被拒绝的AI算力垃圾堆，就是因为他们知道，唯一能维持增长的办法，就是幻想自己可以间接地无限抽干风投和债务市场。

I think the era of hyperscale growth is ending, and the only reason they're so desperately joining the no IT loads, refused AI compute cum dump is because they know that the only way to keep growth going is to hope they can drain the venture capital and debt markets infinitely by proxy.

不知道大家会不会喜欢我刚刚说的那些话。

Don't know if people are gonna like that I just said that.

不过我会继续讲下去。

I'll keep going, though.

总之，现在已经没有其他高增长的市场了。

Anyway, there are no other hyper growth markets left.

这整件事其实就是这么回事。

That's all this is about.

接下来几个月，很多数字还会涨得更高。

So many numbers are going up so much higher in the next few months.

然后到了某个点，它们就不再涨了，接着开始往下跌，有些甚至会归零，甚至变成负数。

Then at some point, they'll stop going higher, and then they'll start going lower, and then some of them will go to zero or even become negative numbers.

因为只有癌症才是永远增长的。

Because the only thing that grows forever is cancer.

现在这一切可能看起来难以置信，因为微软、谷歌和亚马逊的数字还在涨，但这其实只是靠OpenAI和Anthropic在撑着。

This all may seem unthinkable right now because Microsoft, Google, and Amazon numbers are going up, but that's really only happening because of OpenAI and Anthropic.

大多数人没料到会这样，因为这一切都源于对历史上最无厘头、最夸张的营销活动的反击。那些真正的记者和媒体机构，竟然去攻击和骚扰那些敢于质疑权贵的人。

Most didn't see this coming because it has involved pushing back against one of the most gratuitous and excessive marketing campaigns in history involving actual reporters that are real outlets attacking and harassing people for daring to question the powerful.

那些人最终会被追究责任，大家记住我这句话。

And those people will be held accountable, mark my words.

不过咱们还是聊聊真正重要的人，也就是你。

But let's talk about who actually matters, and that's you.

在座的好多人已经陪我走过好几年了，听我把这些观点一点点构建起来。我很感激你们通过 Signal、终端论坛、Reddit 子版块以及邮件给我的支持。

And a lot of you have been here with me for years, hearing me build these arguments, and I appreciate the amount of support I hear from you via Signal, the terminal, and on the subreddit, and email, of course.

你们都很棒，在我遇到困难、别人施压或者对我特别恶劣的时候，是你们让我坚持下来。

You're all awesome, and you keep me going when things are hard, when people are pressing back, when people are genuinely horrible to me.

我不想抱怨。

I don't wanna complain.

我能做世界上最好的工作。

I get to do the best job in the world.

录制这档播客和写通讯稿真的酷毙了。

It fucking rocks to record this podcast and write the newsletter.

这是我做过的最棒的工作。

It's the greatest job I've ever done.

做这件事让我开心得不得了。

I couldn't be happier doing it.

虽然压力很大，但因为有你们，也因为我是真心热爱，所以我一直坚持着。

It's stressful, but I keep going because of all of you and because I really truly enjoy it.

下周我会回来，借用Drill的话说，我会用聪明的方式揭露社会上的所有胡扯。

I'll be back next week and to quote Drill, I'll be calling out all the bullshit of society the smart way.

而且不管多难，我都会免费做这件事。

And against all odds, I'll do it for free.

谢谢你们的收听。

Thank you for listening.

嘿，波特兰迪亚的粉丝们。

Hey, Portlandia fans.

这里是凯莉·布朗斯坦和弗雷德·阿米森。

Carrie Brownstein and Fred Armisen here.

九十年代的梦想，现在以播客的形式重现了。

The dream of the nineties is alive in podcast form.

我们推出了Podlandia，也就是《AO Rewatch》，一档全新的播客节目，我们一起回顾《波特兰迪亚》的每一集，拆解各个小品，深入探讨我们那些标志性的角色，并且揭开幕后制作的神秘面纱。

We're launching Podlandia, AO Rewatch, our brand new podcast where we revisit every episode of Portlandia together, breaking down sketches, going deep on our iconic characters, and pulling back the curtain on how it all got made.

我们还会邀请那些帮助这一切成真的人，包括客串明星、合作伙伴和朋友，比如导演乔纳森·克雷泽尔，市长本人凯尔·麦克拉克伦，传奇音乐人艾米·曼恩，还有更多嘉宾。

And we'll also be joined by the people who helped bring it all to life, Guest stars, collaborators, and friends, including director Jonathan Kreisel, the mayor himself, Kyle McLaughlin, legendary musician Amy Mann, and many more.

凯尔在这里可是全力以赴了。

Kyle is going for it here.

你不仅完全即兴编了词，还即兴唱了一首关于……的歌。

You fully improvised not just words, but a song about

好吧，我以为你们都会写首歌呢。

Well, I thought you're all gonna write a song.

我记得你当时是这么想的。

I remember you thinking that.

听着。

Listen

去Podlandia、AO Rewatch，在iHeartRadio应用、Apple Podcasts，或者你平时听播客的任何地方都能找到。

to Podlandia, AEO rewatch on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.

我是尼克·塔图罗。

I'm Nick Taturo.

你可能通过《纽约重案组》、《最长的一码》，或者斯派克·李的《黑色党徒》认识我。

You probably know me from NYPD Blue, The Longest Yard, or Spike Lee's Black Klansman.

在我的新播客《尼克·塔图罗的快乐配送》里，我会给每位新嘉宾送披萨。

And on my new podcast, Delivering Happiness with Nick Taturo, I deliver pizza to a new guest.

我把这块蛋糕分给了塞斯·罗林斯，他当时还问：你拿我的腰带干嘛？

I've shared a slice with everyone from Seth Rollins What are you doing with my belt?

还有比尔·伯克。

To Bill Burke.

我觉得我从来没遇到过像我这么疯的人。

I don't think I've ever met somebody so exactly out of their mind as I am.

现在，我们还有更多重磅嘉宾要来，包括约翰·多图罗。

And now, we even have more great guests coming up, including the great John Dotturro.

这叫做幸福。

It's called happiness.

传递幸福。

Delivering happiness.

还有很多很多。

And many, many more.

打开免费的iHeartRadio应用，搜索“与尼克·多图罗一起传递幸福”，现在收听。

Open your free iHeartRadio app, search delivering happiness with Nick Dotturro, and listen now.

我是切尔西·汉德勒，来自《亲爱的切尔西》。

This is Chelsea Hamler from Dear Chelsea.

每周，新闻越来越糟，世界越来越疯狂，而雅梅内卡在这里告诉那些该负责的人：你们就是问题所在。

Every week, the news gets worse, the world gets crazier, and Yamaneka is here to tell whoever's responsible, you're the problem.

你知道吗，我刚知道悉尼·斯威尼是谁。

Do you know I just found out who Sydney Sweeney was?

如果他有一堆女人，那我也应该有一堆男人。

If he got a bunch of women, then I should have a bunch of men.

要么做得更好，要么少做点，省得我干那么多活。

Do better or do less so I don't have to do so much.

我是雅曼尼卡，我先走了。

I'm Yamanika, and I'm out.

在雅曼尼卡的《你就是问题》节目里收听

Listen to You're the Problem with Yamanika on

iHeartRadio应用、Apple Podcast，或者任何你收听播客的地方。

the iHeartRadio app, Apple Podcast, or wherever you get your podcasts.

暑期大片回来了。

Summer blockbusters are back.

《奥德赛》和《蜘蛛侠》全新系列强势回归。

The Odyssey and Spider Man, brand new day have taken over series.

节目。

Show.

我是

I'm the

第一个到片场的。

the first one on set.

除了背熟自己的台词，他还能记住片场所有人的名字。

Aside from learning all of his lines, he's also able to memorize everybody's name on set.

下载免费的iHeartRadio应用，搜索《夺宝奇兵》播客，现在收听。

Download the free iHeartRadio app, search Raiders of the Lost podcast, and listen now.

这是iHeart播客。

This is an iHeart podcast.

保证真人。

Guaranteed Human.